Active optical cables have been garnering attention as a transmission medium to serve as an alternative to metal cables. An active optical cable is configured such that it includes (i) a cable containing an optical fiber; and (ii) a pair of connectors respectively provided on both ends of the cable. One of the connectors functions as an optical transmitter. This optical transmitter converts, into an optical signal, a voltage signal externally supplied (e.g., from a data center computer) and then transmits the optical signal. The other connector serves as an optical receiver. This optical receiver converts a received optical signal into a voltage signal and then outputs the voltage signal outside the optical receiver (e.g., to a data center storage device). By making each of the connectors function as both an optical receiver and an optical transmitter, it is possible to realize bidirectional communication using the active optical cable.
FIG. 9 illustrates an optical receiver 2. The optical receiver 2 is a conventional optical receiver that can be used as a connector for an active optical cable. The optical receiver 2 includes (i) a photodetector element 21 that converts an optical signal into an electric current signal, and (ii) a receiving circuit 22 that converts the electric current signal into a voltage signal.
The receiving circuit 22 is configured such that it includes a transimpedance amplifier 22a, differential amplifiers 22b through 22e, a low-pass filter 22f, and an error amplifier 22g. 
The transimpedance amplifier 22a converts, into a voltage signal (single end), an electric current signal outputted from the photodetector element 21. The differential amplifier 22b performs differential amplification of the difference between (i) a voltage signal outputted from the transimpedance amplifier 22a and (ii) a threshold voltage Vth. By performing this differential amplification, the differential amplifier 22b obtains a differential signal consisting of a positive phase signal and a negative phase signal. The group of differential amplifiers 22c through 22e performs differential amplification of the differential signal outputted from the differential amplifier 22b. 
If the output voltage of the transimpedance amplifier 22a is Vtia, then the differential amplifier 22b has a positive phase output voltage V1p which is expressed as V1ocm+a1×(Vtia−Vth)/2, and a negative phase output voltage V1n which is expressed as V1ocm−a1×(Vtia−Vth)/2. Here, V1ocm is an output common mode voltage (a predetermined value) of the differential amplifier 22b, and a1 is a gain (a predetermined value) of the differential amplifier 22b. 
An average value of a high level and a low level of a voltage signal outputted by the transimpedance amplifier 22a is, hereinafter, also referred to as an “average output level of transimpedance amplifier 22a.” In a case where the average output level of the transimpedance amplifier 22a is equal to a threshold voltage Vth, the positive phase signal and the negative phase signal outputted from the differential amplifier 22b have waveforms which become symmetrical to each other with respect to the output common mode voltage V1ocm. However, in a case where the power of the optical signal being received fluctuates and the transimpedance amplifier 22a has an average output level that is not equal to the threshold voltage Vth, the positive phase signal and the negative phase signal have respective waveforms which become asymmetrical to each other with respect to the output common mode voltage V1ocm. This sort of asymmetry causes distortion of a waveform of an output signal of the optical receiver 2.
The low-pass filter 22f and the error amplifier 22g are each a component for avoiding the above problem. The low-pass filter 22f performs smoothing of (i) the positive phase signal outputted from the differential amplifier 22c and (ii) the negative phase signal outputted from the differential amplifier 22c. The error amplifier 22g receives: (i) a smoothed positive phase signal (DC component of the positive phase signal) outputted from the low-pass filter 22f, and (ii) a smoothed negative phase signal (DC component of the negative phase signal) outputted from the low-pass filter 22f. The error amplifier 22g then integrates the difference between the respective values of these two smoothed signals, i.e., an offset voltage of the differential signal outputted from the differential amplifier 22c. A resulting integration value of the offset voltage outputted from the error amplifier 22g is fed back, as a threshold voltage Vth, into a negative phase input of the differential amplifier 22b. 
The integration value of the offset voltage outputted from the error amplifier 22g follows the average output level of the transimpedance amplifier 22a. As a result, even if the power of the optical signal being received fluctuates, the abovementioned distortion problem is avoided.
Patent Literature 1 is an example literature disclosing art that cancels an offset voltage of a differential signal.